Statistical Clues to Social Injustice, A paper from Radstats

« Older   Newer »
  Share  
FionaK
view post Posted on 14/5/2011, 22:04




I am not sure if I am posting this in the right place: it is a paper on statistics and it may be better placed in another part of the forum. But statistics presents itself as a social science and so this placement makes sense to me: even though the thesis points to some problems with stats as any kind of science. Nonetheless the problems of interpretation bedevill an awful lot of what is termed "social science": so much so that it is possible to argue they are utterly misnamed. But then that problem of interpretation is also present in "harder" science and the lines are difficult to draw. So I will stick with this for now

www.radstats.org.uk/no102/Dorling102.pdf

QUOTE
We tend to take the statistics international bodies present to us with far too few grains of salt. Figure 1 shows one such set of numbers. This is how children have been described in the Netherlands in the latest of the PISA international studies of children. This particular study concerned science and maths. Like me you probably don’t know very much about children in the Netherlands. Perhaps there really are 2% with no ability at all, a further 11% who are very limited in their abilities, a further 21% who are ‘barely able’ and 27% who do better than these but can only understand ‘simple concepts’? That’s 61% of all children being dullards.
This is important information if you visit the Netherlands. Look out for all those stupid children. Let’s hope that the Dutch are being careful enough to ensure that none of these young people end up in any positions of power. Fortunately some 2% of their offspring appear to be ‘advanced’ according to these statistics. Obviously they need to carefully ensure that these few are rapidly promoted to positions of responsibility. Those few will need to be supported by some others who can understand their instructions.
Oh Look! There are another 11% beneath them who could fulfil that role, and under them! An even bigger group, some 26% who can be ‘effective’ if properly instructed. Perhaps we can place them between those destined to govern and the 61% who need to be looked after? Perhaps in future generations we could teach children better so that a few more could climb up the ranks (hand out a few more Cuisenaire Rods), but in the meantime we have a big problem in the Netherlands with that 61% incapable of looking after themselves in any intellectual sense. But what about other countries? What about the UK where most people reading this journal live? Have a look at figure 2.

It’s terrible! What have our schools being doing? Our teachers must be awful! More than twice as many young people in Britain have no measurable ability. Hold on though, look to the other end of the table. There’s some hope, some 2.9% of our young people are ‘advanced’. Make’s you proud to be British don’t you think? - more than in both the USA and the Netherlands, clearly we must be doing something right. Maybe it’s our excellent private schooling at the very peak of our system, Eton, Harrow, Westminster, our “very best” comprehensives contributing too, all creating the cream-of-the-cream. Maybe having more at the bottom is a sacrifice we have to make for bringing on excellence at the top?

Lies, damned lies, and statistics: as the man said :)
 
Top
view post Posted on 15/5/2011, 00:12
Avatar

Member

Group:
Administrator
Posts:
756

Status:


I just read the article and got the feeling he didn't get to the point... I appreciate the examples, but it feels like he's just pricking some issues, whimsically. I suspect he's working on a lot of culturally active topics that I am not quite into...?
 
Top
FionaK
view post Posted on 15/5/2011, 00:57




Well as it said it was a talk based on a more detailed paper and previous work in a book. He was illustrating how stats can be misleading. For me the main point to grasp is the one about imposing a normal distribution. That is something which I ran into when I was a student and it was one of those things that just get you into trouble. I argued that it was not shown that things were normally distributed and the professor decided I was an idiot. I was not able to articulate my point.I suppose that if you have not had that rankling for a long time the article is less interesting :) So sorry if it did nothing for you: was helpful to me
 
Top
view post Posted on 4/4/2016, 00:53
Avatar

Member

Group:
Administrator
Posts:
756

Status:


This is relevant: http://fivethirtyeight.com/features/scienc...t-broken/#part1

It has some charts to play with. And it features a story about sending identical data sets to different teams of analysts, and getting widely different statistical conclusions. It's all based on subjective choices on how to interpret this data.
 
Top
3 replies since 14/5/2011, 22:04   66 views
  Share